No photosynthetic down-regulation in sweetgum trees (Liquidambar styraciflua L.) after three years of CO₂ enrichment at the Duke Forest FACE experiment

ABSTRACT

Photosynthetic capacity and leaf properties of sun and shade leaves of overstorey sweetgum trees (Liquidambar styraciflua L.) were compared over the first 3 years of growth in ambient or ambient + 200 μL L⁻¹ CO₂ at the Duke Forest Free Air CO₂ Enrichment (FACE) experiment. We were interested in whether photosynthetic down-regulation to CO₂ occurred in sweetgum trees growing in a forest ecosystem, whether shade leaves down-regulated to a greater extent than sun leaves, and if there was a seasonal component to photosynthetic down-regulation. During June and September of each year, we measured net photosynthesis (A) versus the calculated intercellular CO₂ concentration (C_i) in situ and analysed these response curves using a biochemical model that described the limitations imposed by the amount and activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (Vc_max) and by the rate of ribulose-1,5-bisphosphate (RuBP) regeneration mediated by electron transport (J_max). There was no evidence of photosynthetic down-regulation to CO₂ in either sun or shade leaves of sweetgum trees over the 3 years of measurements. Elevated CO₂ did not significantly affect Vc_max or J_max. The ratio of Vc_max to J_max was relatively constant, averaging 2·12, and was not affected by CO₂ treatment, position in the canopy, or measurement period. Furthermore, CO₂ enrichment did not affect leaf nitrogen per unit leaf area (N_a), chlorophyll or total non-structural carbohydrates of sun or shade leaves. We did, however, find a strong relationship between N_a and the modelled components of photosynthetic capacity, Vc_max and J_max. Our data over the first 3 years of this experiment corroborate observations that trees rooted in the ground may not exhibit symptoms of photosynthetic down-regulation as quickly as tree seedlings growing in pots. There was a strong sustained enhancement of photosynthesis by CO₂ enrichment whereby light-saturated net photosynthesis of sun leaves was stimulated by 63% and light-saturated net photosynthesis of shade leaves was stimulated by 48% when averaged over the 3 years. This study suggests that this CO₂ enhancement of photosynthesis will be sustained in the Duke Forest FACE experiment as long as soil N availability keeps pace with photosynthetic and growth processes.

INTRODUCTION

Forest ecosystems function as an important link between terrestrial and atmospheric portions of the global carbon cycle, since forests cover 43% of the Earth's land surface and account for as much as 70% of the terrestrial net primary productivity (Melillo et al. 1993). Recent models and empirical studies suggest that the temperate forests of North America act as an important sink for atmospheric CO₂ (Wofsy et al.; Tans & Bakwin 1995; Fan et al. 1998). This may in part result from a stimulation of net photosynthesis and growth of forest trees by rising atmospheric CO₂ (Eamus & Jarvis 1989; Poorter 1993; Ceulemans & Mousseau 1994; Gunderson & Wullschleger 1994; Curtis 1996). In a recent meta-analysis of 500 CO₂ enrichment studies using tree species, Curtis & Wang (1998) found that a doubling of CO₂ concentration stimulated leaf level photosynthesis on average by approximately 54%. The magnitude of this response, however, may be modulated by abiotic factors such as irradiance, soil moisture, and soil nutrient availability (Curtis & Wang 1998). Understanding the photosynthetic responses of trees to CO₂ enrichment when growing within the full suite of forest ecosystem processes is essential to assessing the potential of forests to store carbon.

Many predictions of greater forest productivity with increasing CO₂ depend on a sustained increase in the photosynthetic capacity of leaves (Reynolds et al. 1996). A feature of many CO₂ studies, however, is a time-dependent decline in the degree of photosynthetic enhancement by elevated CO₂. For example, in an assessment of studies using 39 tree species, Gunderson & Wullschleger (1994) found that the long-term CO₂ enhancement of photosynthesis was on average 21% lower than the short-term enhancement. This decline in photosynthetic enhancement has been termed ‘photosynthetic down-regulation' and is characterized by a decrease in leaf nitrogen and ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), which leads to the reduction of photosynthetic capacity (Stitt 1991; Long & Drake 1991; Bowes 1993; Sage 1994; Norby et al. 1999). Greater growth rates lead to an increased N demand by plants (Ingestad & Stoy 1982), and many forest ecosystems are N-limited (Vitousek & Howarth 1991). Therefore, this type of photosynthetic down-regulation might be expected to be a common occurrence in forest systems that have a low soil N availability (Radoglou, Aphalo & Jarvis 1992; Tissue, Thomas & Strain 1993; Sage 1994; El Kohen & Mousseau 1994; Curtis et al. 1995), such as a piedmont North Carolina loblolly pine forest. We hypothesized that CO₂ enrichment would cause a time-dependent reduction in photosynthetic capacity of overstorey sweetgum (Liquidambar styraciflua L.) trees in the Duke Forest Free Air CO₂ Enrichment (FACE) experiment and that this down-regulation would be associated with a decline in leaf N. Further, we expected greater photosynthetic down-regulation later in the growing season because of a decline in leaf N.

The light environment in a forest ecosystem is another important factor in determining whether or not photosynthetic down-regulation to CO₂ will occur, as light attenuation through the canopy plays a strong role in the developmental, morphological, and physiological characteristics of leaves (Boardman 1977; Björkman 1981; Ellsworth & Reich 1993). For example, in the Duke Forest FACE experiment, we found that sun leaves of sweetgum trees were thicker and had more N and chlorophyll per unit leaf area than shade leaves, as well as higher light-saturated net photosynthetic rates (Herrick & Thomas 1999). In addition, the CO₂ stimulation of light-saturated photosynthesis of sun leaves of sweetgum trees was greater than the CO₂ stimulation of shade leaves (Herrick & Thomas 1999). Leaf N may be optimally partitioned between the carboxylating and light-harvesting processes of photosynthesis (Evans 1987, 1989). Sage (1994) proposed that photosynthetic acclimation could occur as a consequence of increased atmospheric CO₂ if the capacity of carboxylation exceeds the capacity to regenerate RuBP, thus altering the partitioning of N from non-limiting processes to processes limiting production. Partitioning of N in leaves developed under low irradiance produces excess carboxylation capacity, and it has been proposed that concurrent CO₂ enrichment and low irradiance should produce a stronger N reallocation response than with either factor alone (DeLucia & Thomas 2000). Thus, we predicted that shade leaves of sweetgum trees would show greater photosynthetic down-regulation to CO₂ than sun leaves.

The majority of CO₂ enrichment studies on tree species have been confined to container-grown seedlings and saplings in controlled environments. Pot size has a pronounced effect on photosynthetic down-regulation to CO₂ (Thomas & Strain 1991), and down-regulation is common in studies using small pots (Gunderson & Wullschleger 1994; Curtis & Wang 1998). The reduction in photosynthetic capacity in pot studies has been postulated to result from restricted root growth which causes a feedback mechanism triggered by sugar or starch accumulation in the leaf with the end result of biochemical down-regulation of Rubisco activity (Thomas & Strain 1991; Stitt 1991). In contrast, a number of studies of field-grown trees, where rooting volume is not restricted, have observed little or no evidence of photosynthetic down-regulation due to CO₂ enrichment (Gunderson, Norby & Wullschleger 1993; Ellsworth et al. 1995; Vogel & Curtis 1995; Scaracia-Mugnozza et al. 1996; Myers, Thomas & DeLucia 1999; Medlyn et al. 1999). Indeed, Curtis & Wang (1998) in their meta-analysis found little evidence of photosynthetic down-regulation in trees when plants were grown in containers with a volume larger than 0·5 L.

Thus, many questions remain as to whether or not the initial stimulation of photosynthesis observed for seedlings and saplings with a doubling of CO₂ will be sustained for trees growing in a forest ecosystem under natural conditions. The objective of our study was to assess the photosynthetic capacity of sweetgum trees during the first three years at the Duke Forest FACE experiment. We were specifically interested in whether photosynthetic down-regulation occurred in sweetgum trees growing in a low nutrient forest ecosystem, whether shade leaves acclimated to a greater extent than sun leaves, and whether there was a seasonal component to photosynthetic down-regulation. We were also interested in whether a decline in photosynthetic capacity was associated with a decline in leaf N, reallocation of N within the photosynthetic apparatus, or an increase in leaf carbohydrates. The Duke Forest FACE study is situated in a North Carolina piedmont forest dominated by loblolly pine trees (Pinus taeda L.) with naturally established sweetgum trees as the dominant deciduous tree species. Sweetgum is an early successional tree species that commonly invades broomsedge (Andropogon virginicus L.) fields in piedmont North Carolina during the course of secondary succession (Oosting 1942). Photosynthetic capacity of sweetgum sun and shade leaves was assessed by measuring light saturated net photosynthesis (A) versus the calculated intercellular CO₂ concentration (C_i) and analysing these response curves using a biochemical model that describes the limitations imposed by the amount and activity of Rubisco and by the rate of ribulose-1,5-bisphosphate (RUBP) regeneration mediated by electron transport (Harley & Sharkey 1991). The degree and potential causes of photosynthetic down-regulation of sweetgum leaves were examined using the electron transport/carboxylation ratio modeled from the A–C_i response curves (Medlyn 1996).

MATERIALS AND METHODS

Duke Forest Free-Air CO₂ Enrichment (FACE) experiment

The Duke Forest FACE experiment is located in a Pinus taeda plantation in the Blackwood division of the Duke Forest (35°97′ N 79°09′ W). No management measures have been taken to prevent the growth of other tree species since the current plantation was established in 1983 after the regenerating forest was clear-cut in 1979. As a result, the forest is dominated by loblolly pine (Pinus taeda 1733 stems ha⁻¹), but there are significant numbers of sweetgum (Liquidambar styraciflua, 620 stems ha⁻¹) and yellow poplar (Liriodendron tulipifera L., 68 stems ha⁻¹) as well as other hardwood species in the overstorey and the understorey. The loblolly pine trees were approximately 11 m tall in 1996 and 14 m tall in 1999. Sweetgum trees in the forest overstorey were 7–11 m in height in 1997 and 10–14 m in height in 1999. This forest is growing on a nutrient-poor, clay-rich loam soil that is typical of many upland areas in southeastern USA.

Within this forest, six 30-m-diameter experimental circular plots were established. Three of these FACE rings are replicate CO₂ treatments and the remaining three plots are ambient experimental controls. Each FACE ring consists of 32 vertical pipes that extend from the forest floor through the forest canopy. In the elevated treatment FACE rings, these pipes deliver a controlled amount of CO₂ throughout the entire forest volume, with a target CO₂ concentration of ambient + 200 μL L⁻¹. Three control rings receive the same volume of air to replicate any micrometeorological effects on the forest that occurs during the operation of the Duke Forest FACE facility. The Duke Forest FACE experiment began full operation during August 1996, and since then, the CO₂ treatment has been continuously applied 24 h d⁻¹ except when the air temperature was below 5 °C for more than an hour. During the first 3 years of the experiment, the daytime average CO₂ concentration was 573 μL L⁻¹ in the elevated treatment rings and 378 μL L⁻¹ in the ambient control rings. To control for topographic variation (approximately 5 m) and potential gradients in site fertility between rings, the three control and three elevated CO₂ rings were arranged in a complete block design (three pairs).

Gas exchange and leaf chemistry

Measurements were made twice during the growing season, within a single week period around 25 June and 1 September of 1997, 1998 and 1999. June measurements were made approximately 68 d after leaf initiation and the September measurements were made approximately 50 d before leaves began to senesce. Minimum air temperatures during June and September sample periods ranged from 15 to 20 °C and maximum temperatures ranged from 32 to 35 °C. A drought occurred during late July, August and early September of both 1997 (Ellsworth 1999) and 1998 (Fig. 1). During 1997, soil moisture averaged 19·6% in late June and 13·8% during early September (Fig. 1; K. Schäfer, Duke University, unpublished results). In 1998, soil moisture averaged 24·4% in late June and 15·0% during early September. During 1999, soil moisture averaged 24·3% in late June and 20·9% during early September (Fig. 1).

Figure 1. Environmental conditions at the Duke Forest FACE experiment during the first 3 years of treatment. Parameters include mean monthly CO₂ concentrations based on daily averages during the daylight hours, total monthly precipitation, average monthly maximum and minimum temperature and soil moisture. Arrows indicate the periods during which photosynthetic parameters of sun and shade leaves of Liquidambar styraciflua (L.) trees were measured.

Two overstorey sweetgum trees (7–11 m in height in 1997) were selected in each FACE ring based on the proximity of trees to areas accessible from portable hydraulic lifts. All of these trees had leaves exposed to full sunlight at the top of the crown and deep shade at the bottom of the crown. The same trees and leaf positions were used in Herrick & Thomas (1999). Diffuse irradiance for the shade leaves was approximately 50 μmol m⁻² s⁻¹ punctuated by intermittent sunflecks. Sun leaf irradiance was typically saturating and varied between 1100 and 1400 μmol m⁻² s⁻¹ during midday on sunny days (Herrick & Thomas 1999).

Light-saturated net photosynthesis (A) was examined in situ on one sun leaf and one shade leaf from each tree using an open flow infrared gas analyser with an attached LED light source (LI6400, Li-Cor, Inc. Lincoln, NE, USA). The response of A to calculated intercellular CO₂ (C_i) was measured over a range of 10 external CO₂ partial pressures (C_a) from approximately 50–1200 μL L⁻¹ (Farquhar & Sharkey 1982). Measurements were made with a constant saturating irradiance of 1400 μmol m⁻² s⁻¹ photon flux density. Preliminary trials indicated that photosynthetic rates reached steady state within 3 min following an increase in C_a. Fully expanded leaves at least two to three weeks old, determined by a concurrent study of sweetgum leaf demography (Herrick, unpublished results), were used. Gas exchange measurements were restricted to the hours between 1000 and 1500 h on sunny days to minimize diurnal effects on photosynthesis. Leaf temperatures were not controlled during measurements and did not differ between the treatments. Leaf temperatures averaged 34·1 ± 0·4 °C in June 1997, 32·3 ± 0·8 °C in September 1997, 32·3 ± 0·4 °C in June 1998, 35·0 ± 0·4 °C in September 1998, 27·5 ± 1·0 °C in June 1999 and 30·3 ± 0·5 °C in September 1999. Vapour pressure deficit averaged 1·95 ± 0·17 and did not differ between CO₂ treatment, sample period or canopy position. Trees in one blocked pair of rings were measured each day so that slight differences in daily weather conditions could be included in the block effect in the analysis of variance.

The A–C_i curves were analysed using a biochemically based model (Farquhar et al. 1980; Harley et al. 1992) to determine the light-saturated rates of carboxylation (Vc_max) and the rate of RuBP regeneration mediated by electron transport (J_max) using the kinetic parameters of Wullschleger (1993). J_max was calculated using C_a at or below 1200 μL L⁻¹ to avoid the region of the A–C_i curve where triose phosphate use limitations may occur (Harley & Sharkey 1991). Net photosynthesis at growth CO₂ (A_net) was obtained from A–C_i curves where C_a equalled 360 μL L⁻¹ for control leaves and 560 μL L⁻¹ for CO₂-enriched leaves.

Leaf properties

After gas exchange measurements were completed, each leaf was harvested and assayed for N, chlorophyll and non-structural carbohydrates. Leaf mass per unit area (LM_a, g m⁻²) was calculated by measuring the dry mass of leaf discs of a known area. Leaf tissue for carbohydrate and N analyses was frozen in liquid N₂, stored at –80 °C, and later dried to a constant mass at 65 °C.

N was assayed following Dumas combustion with a Carlo Erba CNS autoanalyzer (Fisons Instr., Milan, Italy). Least squared linear regression was used to plot Vc_max and J_max as a function of N per unit leaf area (N_a), pooling measurements from sun and shade leaves from 1998 and 1999 to get a range of N_a values.

Leaf soluble sugar and starch concentrations were assayed spectrophotometrically as described by Thomas & Griffin (1994). Tissue for chlorophyll analysis was immediately stored on ice and transported to the laboratory where chlorophyll was extracted from leaf tissue with N,N-dimethylformamide and analysed spectrophotometrically according to the methods of Porra, Thompson & Kriedman (1989). Leaf properties were measured on gas exchange leaves harvested during 1998 and 1999. Due to technical reasons, measurements could not be made on our specific gas exchange leaves in 1997. However, chemistry data was collected for similar leaves harvested during the two measurement periods in 1997 and can be found in Herrick & Thomas (1999).

Data analysis

Gas exchange parameters and leaf properties were analysed using a repeated measures analysis of variance (ANOVA) model with CO₂ treatment, leaf position, month, year and blocked ring pair as main effects (Data Desk 1997). Individual rings (n = 3) were considered replicates for the purposes of statistical analysis. Post hoc comparison of parameter means was performed using the Bonferroni multiple comparisons test (Data Desk 1997). Parameters were considered significantly different at the 0·1 probability level. Analysis of covariance (ANCOVA) and least squares linear regression were used to analyse the relationship of leaf N_a to Vc_max and J_max; a significant treatment–N interaction would indicate that the slopes were significantly different between the control and treatment CO₂ rings.

RESULTS

Net photosynthesis

Elevated CO₂ strongly stimulated light-saturated net photosynthetic rates (A_net) of sun and shade leaves of sweetgum trees throughout the first 3 years of the Duke Forest FACE experiment (P = 0·002; Fig. 2). However, A_net of sun leaves was stimulated to a greater degree by CO₂ enrichment than was A_net of shade leaves (CO₂× leaf position; P = 0·015). Averaged across all sample periods, A_net of sun leaves was stimulated by 63% when trees were grown under elevated CO₂ whereas A_net of shade leaves was stimulated by 48%. Net photosynthesis of sun leaves was much greater than A_net of shade leaves and averaged 83% greater than shade leaves across the three years of the experiment (P = 0·015).

Figure 2. Light saturated net photosynthesis (A_net) of sun and shade leaves of overstorey Liquidambar styraciflua (L.) trees growing at the Duke Forest FACE experiment under ambient (closed symbols) and elevated (open symbols) CO₂. Gas exchange was measured near 25 June and 1 September of each year. Measurements were made at saturating irradiance and were restricted to the hours between 1000 and 1500 h on sunny days. Each point is the mean of three rings (± SE) for each CO₂ treatment.

We found a month × year interaction on A_net of sweetgum leaves (P = 0·059). A_net was 22% (P = 0·073) and 36% (P = 0·013) greater in June 1997 and 1998, respectively, than the September sample period of those years. On the other hand, A_net of sweetgum leaves was not significantly different during June and September 1999. We also found a leaf position × year interaction (P = 0·024), whereby there was an 18% increase in sun leaf A_net in 1999 compared with 1997 and 1998. There was no effect of year on A_net of shade sweetgum leaves. In addition, we found no CO₂× year interactions on A_net of sun or shade sweetgum leaves.

Photosynthetic capacity

The relationships between A and C_i for sun and shade leaves of sweetgum trees in Duke Forest FACE are shown in Figs 3 and 4. Two potential biochemical limitations to photosynthesis, the light-saturated rates of carboxylation (Vc_max) and the rate of RuBP regeneration mediated by electron transport (J_max), were modelled from the A–C_i curves. CO₂ enrichment did not have a significant effect on Vc_max or J_max when each was averaged across all sample periods (Fig. 5). Vc_max (P = 0·007) and J_max (P = 0·015) of sun leaves were 80% greater than shade leaves (Fig. 5). Vc_max and J_max were not significantly different between June and September of each year, nor were they significantly different between years. In addition, there were no significant CO₂ treatment × year interactions on Vc_max and J_max. The ratio of J_max to Vc_max was not affected by CO₂ treatment, leaf position, month or year of measurement, and thus was relatively constant, averaging 2·12 (± 0·09 SE).

Figure 3. The response of net photosynthesis to the calculated intercellular CO₂ concentration for sun leaves of overstorey Liquidambar styraciflua (L.) trees grown under ambient (~360 mL L⁻¹; ● and solid lines) and ambient plus 200 μL L⁻¹ CO₂ (○ and dashed lines). These responses were measured using saturating irradiance during the first 3 years of treatment at the Duke Forest FACE experiment. Data were fit using non-linear least squares with the equation: A_net = A_max[1 – (1 –α/A_max)^{(1 −Ci/Γ)}]; where A_max = A_net at CO₂ saturation, α = y-intercept, and Γ = CO₂ compensation point.

Figure 4. The response of net photosynthesis to the calculated intercellular CO₂ concentration for shade leaves of overstorey Liquidambar styraciflua (L.) trees grown under ambient (~360 mL L⁻¹; ● and solid lines) and ambient plus 200 μL L⁻¹ CO₂ (○ and dashed lines). These responses were measured using saturating irradiance during the first three years of treatment at the Duke Forest FACE experiment. Data were fit using non-linear least squares with the equation: A_net = A_max[1 – (1 –α/A_max)^{(1 −Ci/Γ)}]; where A_max = A_net at CO₂ saturation, α = y-intercept, and Γ = CO₂ compensation point.

Figure 5. Photosynthetic parameters, Vc_max (maximum rate of carboxylation), J_max (maximum rate of electron transport), and the ratio J_max : Vc_max modelled from A–C_i curves measured on sun and shade leaves of overstorey Liquidambar styraciflua (L.) trees growing at the Duke Forest FACE experiment under ambient (closed symbols) and elevated (open symbols) CO₂. Gas exchange was measured near 25 June and 1 September of each year. Measurements were made at saturating irradiance and were restricted to the hours between 1000 and 1500 hours on sunny days. Each point is the mean of three rings (± SE) for each CO₂ treatment.

We found a strong relationship between N_a and the modelled photosynthetic parameters (Fig. 6). The slopes for all the regressions of Vc_max and J_max as a function of N_a were significantly greater than zero (P < 0·01). Analysis of covariance indicated that the slopes of N_a versus Vc_max and J_max were not affected by growth CO₂ treatment.

Figure 6. The relationships between modelled photosynthetic parameters, Vc_max (maximum rate of carboxylation) and J_max (maximum rate of electron transport), and leaf N on an area basis (N_a) from overstorey Liquidambar styraciflua (L.) trees growing at the Duke Forest FACE experiment under ambient (● and solid line) and elevated (○ and dashed line) CO₂. Data from 1998 and 1999 and both sun and shade crown positions were used to increase the range of N_a. Equations for the regression lines are given on the figure.

Leaf properties

Elevated CO₂ did not significantly affect leaf N on a mass basis (N_m) or on a unit leaf area basis (N_a) over the course of our measurements (Table 1). However, leaf mass per unit area (LM_a) was greater under elevated CO₂ treatment (P = 0·004) and elevated CO₂ enhanced LM_a of sun leaves to a greater degree than shade leaves (CO₂× leaf position; P = 0·062). LM_a of sun leaves was 23% greater under CO₂ enrichment and LM_a of shade leaves was 8% greater under CO₂ enrichment.

Table 1.  Chemical and morphological properties of sun and shade leaves from overstorey Liquidambar styraciflua (L.) leaves grown in a forest ecosystem under ambient and elevated CO₂. The properties include leaf N on a mass basis (N_m), leaf N on an area basis (N_a), chlorophyll on a mass basis (Chl_m), chlorophyll on an area basis (Chl_a) and leaf mass per unit area (LM_a). Leaves were collected immediately following gas exchange measurements during June and September of 1998 and 1999. Each value is the mean of three rings (± SE) from both CO₂ treatments, ignoring block effects. Means within a row or column followed by the same letter are not statistically different according to Bonferroni's post hoc test (P > 0·1)

Date of measurement:
Position	CO2 treatment	June 1998	Sept. 1998	June 1999	Sept. 1999
Ambient 39·14 ± 2·25^c 39·30 ± 2·65^c 37·93 ± 1·62^c 38·79 ± 2·79^c
Nm (mg g^−1)
Sun	Elevated	17·64 ± 0·91^a	16·44 ± 0·33^a	15·48 ± 0·51^a	17·34 ± 0·67^ac
	Ambient	19·40 ± 0·98^ab	18·39 ± 1·45^a	17·65 ± 0·34^a	18·37 ± 1·08^ac
Shade	Elevated	17·31 ± 0·68^a	16·31 ± 0·27^a	17·06 ± 0·08^a	16·04 ± 0·56^ac
	Ambient	20·65 ± 0·74^b	18·56 ± 0·91^abc	20·06 ± 1·50^bc	17·89 ± 1·43^c
Na (g m^−2)
Sun	Elevated	1·56 ± 0·03^ac	1·44 ± 0·21^ac	1·28 ± 0·10^a	1·73 ± 0·05^c
	Ambient	1·23 ± 0·06^a	1·44 ± 0·18^a	1·21 ± 0·10^a	1·48 ± 0·12^ac
Shade	Elevated	0·70 ± 0·01^b	0·73 ± 0·02^b	0·72 ± 0·04^b	0·65 ± 0·03^b
	Ambient	0·81 ± 0·03^b	0·73 ± 0·04^b	0·75 ± 0·03^b	0·69 ± 0·02^b
Chlm (mg g^−1)
Sun	Elevated	5·49 ± 0·61^a	5·34 ± 0·61^a	6·24 ± 0·75^a	5·54 ± 1·43^a
	Ambient	6·96 ± 0·27^ab	8·07 ± 1·60^ab	8·11 ± 1·79^ab	9·28 ± 1·37^ab
Shade	Elevated	11·83 ± 1·49^b	9·04 ± 0·53^ab	11·31 ± 0·63^ab	12·72 ± 1·03^b
	Ambient	11·49 ± 1·70^b	12·07 ± 1·09^b	13·13 ± 2·38^b	14·02 ± 3·53^b
Chla (g m^−2)
Sun	Elevated	0·48 ± 0·02^a	0·47 ± 0·11^a	0·50 ± 0·05^a	0·58 ± 0·10^ab
	Ambient	0·44 ± 0·05^a	0·61 ± 0·04^ab	0·54 ± 0·07^ab	0·71 ± 0·07^b
Shade	Elevated	0·47 ± 0·05^a	0·40 ± 0·02^a	0·47 ± 0·01^a	0·51 ± 0·02^ab
	Ambient	0·44 ± 0·04^a	0·47 ± 0·03^a	0·50 ± 0·06^a	0·53 ± 0·12^ab
LMa (g m^−2)
Sun	Elevated	89·69 ± 5·94^a	87·91 ± 13·08^a	83·69 ± 4·78^a	100·12 ± 5·88^a
	Ambient	63·49 ± 4·96^b	78·57 ± 9·63^ab	69·53 ± 7·10^ab	80·59 ± 6·43^ab
Shade	Elevated	40·39 ± 2·26^c	44·48 ± 1·39^c	42·00 ± 2·50^c	40·23 ± 1·56^c

Sun and shade leaves of sweetgum trees did not differ in N_m. Expressed as N_a, however, sun leaves had twice the N of shaded leaves (P = 0·008). LM_a of sun leaves was also twice that of shade leaves at all measurement periods (P = 0·003, Table 1). N_m (P = 0·012), N_a (P = 0·065), and LM_a (P = 0·032) differed seasonally, whereby N_m was greater in June than in September, but N_a and LM_a were greater in September than in June. We found no effect of year or CO₂× year interactions on N_m, N_a or LM_a of sweetgum leaves.

Elevated CO₂ had no significant effects on leaf chlorophyll on a mass basis (Chl_m) or on a leaf area basis (Chl_a; Table 1). Shade leaves of sweetgum trees had 77% greater Chl_m than sun leaves (P = 0·037). Chl_a did not differ between sun and shade leaves. Chl_m and Chl_a did not change between June and September measurements or between years, and there were no significant CO₂× year interactions on leaf chlorophyll.

There was no evidence of CO₂ enrichment effects on the concentration of leaf total non-structural carbohydrates (Table 2). Sun leaves had greater soluble sugar concentrations than shade leaves (P = 0·005; Table 2). On the other hand, starch concentrations of sun and shade sweetgum leaves were not different. Leaf sugar concentrations and starch concentrations did not change between June and September sample periods. Leaf sugar concentrations were not affected by year, but leaves contained 16% greater starch concentrations in 1998 than in 1999 (P = 0·078). We found no CO₂× year interactions on leaf non-structural carbohydrate concentrations.

Table 2.  Non-structural carbohydrates of sun and shade leaves from overstorey Liquidambar styraciflua (L.) leaves grown in a forest ecosystem under ambient and elevated CO₂. Leaves were collected immediately following gas exchange measurements during June and September of 1998 and 1999. Each value is the mean of three rings (± SE) from both CO₂ treatments, ignoring block effects. Means within a row or column followed by the same letter are not statistically different according to Bonferroni's post hoc test (P > 0·1)

Date of measurement:
Position	CO2 treatment	June 1998	Sept. 1998	June 1999	Sept. 1999
Ambient 150·6 ± 8·1^b 126·0 ± 3·1^ab 114·9 ± 5·2^a 113·2 ± 2·8^a
Soluble sugar (mg g^−1)
Sun	Elevated	69·0 ± 2·7^a	63·2 ± 3·6^a	55·7 ± 4·3^a	68·9 ± 7·0^a
	Ambient	60·7 ± 12·3^a	63·2 ± 1·0^a	52·7 ± 3·4^a	66·1 ± 8·0^a
Shade	Elevated	52·5 ± 6·0^a	51·9 ± 5·2^a	46·9 ± 4·6^a	54·6 ± 2·6^a
	Ambient	50·1 ± 5·7^a	56·1 ± 3·2^a	48·7 ± 1·1^a	52·2 ± 4·9^a
Starch (mg g^−1)
Sun	Elevated	128·1 ± 10·6^ab	146·7 ± 38·6^a	112·8 ± 7·4^a	109·7 ± 3·5^a
	Ambient	120·5 ± 13·5^ab	131·2 ± 5·8^a	118·4 ± 7·5^a	116·2 ± 6·1^a
Shade	Elevated	118·1 ± 12·3^a	141·5 ± 9·9^a	119·5 ± 3·9^a	113·5 ± 1·5^a

Block effects

In our experiment, there was a significant block paired ring effect for A_net, Vc_max, J_max, N_m, N_a, LM_a, starch and leaf chlorophyll concentration. In all of these parameters except for N_m, the pair of rings that the parameter was measured in did not alter the effect of CO₂ treatment. This was indicated by a non-significant CO₂ treatment × blocked ring pair interaction. On the other hand, there was a significant CO₂ treatment × blocked ring pair interaction (P = 0·001) on leaf N_m. Across all ring pairs and sample periods, N_m averaged 13% less under elevated CO₂ compared to control rings, but this was not a statistically significant difference (P = 0·170). The lack of a significant effect of CO₂ on N_m may have been a result of the variance in response between blocked pairs of rings (CO₂ treatment × blocked ring pair, P = 0·001).

DISCUSSION

Elevated atmospheric CO₂ typically stimulates net photosynthetic rates of C₃ plants, and many models predict greater forest productivity with increasing CO₂. This greater forest productivity depends on the assumption of a sustained photosynthetic enhancement (Reynolds et al. 1996). In many studies, however, a time-dependent decline in the degree of photosynthetic enhancement by elevated CO₂, termed photosynthetic down-regulation, has been observed (Cure & Acock 1986; Gunderson & Wullschleger 1994). We found no evidence of photosynthetic down-regulation in sun or shade leaves of sweetgum trees over the first 3 years of the Duke Forest FACE experiment. There were no time-dependent changes in the stimulation of light-saturated A_net by CO₂ enrichment. In addition, we found no CO₂-dependent effects on leaf photosynthetic capacity or leaf N and chlorophyll on a leaf area basis. Thus, our results do not support the hypothesis that CO₂ enrichment induces photosynthetic down-regulation associated with a decline or reallocation of leaf N. Our data over the first 3 years of the Duke Forest FACE experiment corroborate observations that trees rooted in the ground may not exhibit symptoms of photosynthetic down-regulation as quickly as tree seedlings growing in pots. (Curtis & Wang 1998; Norby et al. 1999; Medlyn et al. 1999).

Photosynthetic down-regulation to elevated CO₂ is often associated with a decline in leaf N on a mass basis (N_m) (Stitt 1991; McGuire et al. 1997; Peterson et al. 1999; Medlyn et al. 1999) and may be more prevalent with low soil N availability than with high N availability (Tissue et al. 1993; El Kohen & Mousseau 1994; Thomas, Lewis & Strain 1994; Curtis et al. 1995; Curtis et al. 2000; Murray et al. 2000). Although growth in elevated CO₂ does not uniformly reduce leaf N in every species examined, N_m is reduced by 16% when averaged across a large number of species (Curtis & Wang 1998). In this study and a previous study with sweetgum (Herrick & Thomas 1999), we found no effects of CO₂ enrichment on N_m or N_a of sun and shade leaves. There were no time-dependent changes in leaf N in low or high CO₂ rings over the three consecutive years of the Duke Forest FACE experiment. Average N_m of sun and shade leaves of sweetgum trees ranged from 15·5 to 20·7 mg g⁻¹ (Table 1), a surprisingly high amount given that the average N_m of forest-grown sweetgum trees is about 15·1 mg g⁻¹ (Blinn & Buckner 1989) and piedmont forests, such as the Duke Forest, are considered to be N-limited forest ecosystems (Adrian Finzi, personal communication). Indeed, Myers et al. (1999) found N_m in loblolly pine needles collected at the Duke Forest FACE experiment were at less than optimal level. Thus, our data indicate that sweetgum trees in the Duke Forest FACE experiment may not be N-limited, and the lack of observed photosynthetic down-regulation in these trees may be related to this result. We did, however, find a strong positive relationship between N_a of sweetgum trees and the modelled components of photosynthetic capacity, Vc_max and J_max (Fig. 6; Medlyn et al. 1999), suggesting that photosynthetic capacity will be affected by any change in N_a as the Duke Forest FACE experiment continues.

Sage (1994) proposed the hypothesis that photosynthetic acclimation to elevated CO₂ may be related to a reoptimization of N within the leaf from carboxylation processes to light-harvesting processes. The ratio J_max : Vc_max represents the balance between the rate of RuBP regeneration via electron transport and the rate of carboxylation by Rubisco. Medlyn (1996) predicted in a theoretical analysis that as growth CO₂ concentration doubles, a reallocation of leaf N from carboxylation processes to light-harvesting components should increase J_max : Vc_max by 40%. Medlyn (1996), however, did not find support for this prediction using data from published CO₂ enrichment studies and subsequently found no evidence of a shift in J_max : Vc_max in a meta-analysis of 15 field-based elevated CO₂ experiments using European tree species (Medlyn et al. 1999). We estimated Vc_max and J_max from 144 A–C_i response curves measured on sun and shade leaves of overstorey sweetgum trees during June and September over the first 3 years of the Duke Forest FACE experiment and found that J_max : Vc_max remained remarkably uniform between CO₂ treatments, sample dates, and sun and shade leaves (Fig. 5). This suggests that there is a strong regulation of photosynthetic capacity of sweetgum leaves, maintaining a constant relationship between carboxylation and light-harvesting capacities over a wide range of environmental conditions.

Specifically, we hypothesized that shade leaves would acclimate to elevated CO₂ to a greater extent than sun leaves by increasing J_max : Vc_max, since shade leaves are light-limited and have excess carboxylation capacity (DeLucia & Thomas 2000). Shade leaves in our study received about 19% of the daily integrated photosynthetic photon flux density that sun leaves received on a typical sunny midsummer day (Herrick & Thomas 1999). These light environments in Duke Forest created predictable differences between sun and shade leaves (Boardman 1977; Björkman 1981), such that shade leaves had about 45% lower A_net (Fig. 2), as well as about 48% lower LM_a and N_a than sun leaves (Table 1). We also found that Vc_max and J_max was about 44% lower in shade leaves than in sun leaves (Fig. 5). However, since we did not find any evidence of photosynthetic down-regulation to CO₂ in either type of sweetgum leaf, our data do not support our hypothesis on differential down-regulation between sun and shade leaves. In other studies, however, elevated CO₂ was found to stimulate an increase in light-harvesting-complex/Rubisco ratio in shaded lower leaves to a greater extent than in the uppermost leaves of wheat plants (Osborne et al. 1998) and to reduce Rubisco content in leaves of strawberry plants growing in a forest understorey by 35% (Osborne et al. 1997). In addition, our data on shade leaves appears to contradict the results from a study using several understorey tree species at the Duke Forest FACE experiment where small increases in J_max : Vc_max suggested that CO₂ enrichment increased the efficiency with which sunflecks were used by the shaded understorey trees (DeLucia & Thomas 2000).

Several studies have shown a seasonal component with regard to photosynthetic down-regulation to CO₂ where Rubisco content declines in the latter part of the growing season (El Kohen & Mousseau 1994; Curtis et al. 1995; Lewis, Tissue & Strain 1996; Rey & Jarvis 1998; Medlyn et al. 1999). We examined this possibility in sweetgum trees by making measurements in early summer and late summer, but we found no seasonal differences in photosynthetic capacity due to CO₂ treatment during the three years of the experiment (Fig. 5). A_net of sun and shade leaves of sweetgum trees were lower in September than June 1997 (17%) and 1998 (26%), but not in 1999, reflecting reductions in soil moisture between June and September in 1997 (43%) and 1998 (63%), but not in 1999. Since neither Vc_max nor J_max of sweetgum leaves showed a seasonal effect, the reduction of A_net between June and September in these 2 years was simply a result of reduced stomatal conductance (data not shown).

Studies using potted plants have often reported reductions in photosynthetic capacity under elevated CO₂ (Gunderson & Wullschleger 1994; Curtis & Wang 1998), including one study where sweetgum seedlings grown for 14 months in elevated CO₂ showed a 24% reduction in Vc_max and J_max compared to ambient CO₂ (Fetcher et al. 1988; Gunderson & Wullschleger 1994). Internal source/sink imbalances produced by CO₂ enrichment have been implicated in photosynthetic down-regulation (Stitt 1991), and small rooting volumes may exacerbate this imbalance (Thomas & Strain 1991). One symptom of a source/sink imbalance of plants growing in elevated CO₂ is the accumulation of non-structural carbohydrates in leaf tissue (Thomas & Strain 1991; Webber, Nie & Long 1994; Cheng, Moore & Seeman 1998), signifying that the rate of photosynthesis is proceeding faster than the rate of photosynthate use. It has been proposed that the accumulation of nonstructural carbohydrates in leaf mesophyll cells leads to a reduction in production of Rubisco (Krapp et al. 1993; Cheng et al. 1998). We did not find any increases in starch or soluble sugars in the sun or shade leaves of sweetgum trees grown under elevated CO₂ during the 3 years of treatment (Table 2), which indicates that the sweetgum trees in the Duke Forest FACE experiment are able to accommodate the extra carbon being fixed as a result of CO₂ enrichment. This contention is supported by other studies at the Duke Forest FACE experiment, where DeLucia et al. (1999) reported that elevated CO₂ stimulated above-ground net primary production of the forest by 12 and 25% during 1997 and 1998, respectively, and Matamala & Schlesinger (2000) reported that elevated CO₂ increased live fine root biomass by 86%.

In conclusion, we found a strong sustained photosynthetic enhancement of sweetgum trees by CO₂ enrichment and no evidence of photosynthetic down-regulation over the first 3 years of the Duke Forest FACE experiment. Averaged across all measurements, light-saturated A_net of sun leaves was stimulated by 63% by elevated CO₂, whereas light-saturated A_net of shade leaves was stimulated by 48%. This was a large enhancement of photosynthesis given that our treatment was 1·5 × ambient CO₂ and the enhancement averaged across many studies using a 2 × ambient treatment was 44 to 54% (Gunderson & Wullschleger 1994; Curtis 1996). Our data, however, are consistent with the CO₂ enhancement of photosynthesis of tree species reported in other studies from the Duke Forest FACE experiment using both sun leaves (Myers et al. 1999; Ellsworth 1999) and shade leaves (DeLucia & Thomas 2000). With no evidence of photosynthetic down-regulation of overstorey sweetgum trees, our study suggests that this large CO₂ enhancement of photosynthesis will be sustained in the Duke Forest FACE experiment as long as soil N availability keeps pace with photosynthetic and growth processes.

ACKNOWLEDGMENTS

We greatly appreciate the soil moisture data provided by Karina Schäfer. Statistical assistance was provided by Dr Don Burdick of the Duke University Institute of Statistics and Decision Sciences. Dr Stephen Long provided the model to estimate the biochemical parameters from leaf gas exchange. We are grateful to Nick Zegre and Jeff Pippen for assistance in data collection and Dr David Myers and Dr Jim Lewis for guidance on the technical aspects of this project. We thank Walter Cybulski, Jon Wallace, and three anonymous reviewers for comments on an earlier draft of this manuscript. We acknowledge the Duke Forest FACE experiment supported by the US Department of Energy. In addition, we appreciate the hospitality and office space provided by the Duke University Phytotron. This research was supported through the NSF/DOE/NASA/USDA/EPA/NOAA Interagency Program on Terrestrial Ecology and Global Change (TECO) and a U.S. Department of Energy PER Grant DE-FG02–95ER62124.